Skip to content

HBASE-28204 : Canary can take lot more time If any region (except the first region) starts with delete markers #5522

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
shahrs87 merged 1 commit into from
Nov 15, 2023
Merged

HBASE-28204 : Canary can take lot more time If any region (except the first region) starts with delete markers #5522

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 22 additions & 22 deletions hbase-server/src/main/java/org/apache/hadoop/hbase/tool/CanaryTool.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -510,38 +510,38 @@ public Void call() {

private Void readColumnFamily(Table table, ColumnFamilyDescriptor column) {
byte[] startKey = null;
Get get = null;
Scan scan = null;
Scan scan = new Scan();
ResultScanner rs = null;
StopWatch stopWatch = new StopWatch();
startKey = region.getStartKey();
// Can't do a get on empty start row so do a Scan of first element if any instead.
if (startKey.length > 0) {
Copy link
Contributor

@d-c-manning d-c-manning Nov 15, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

is there any benefit to retaining the Get path? I legitimately don't know.

If there is, it seems like it could be possible to make a change where we use Get only if (startKey.length > 0 && !rawScanEnabled)

But if Get is unnecessary, then your proposal is good, because it reduces the complexity anyway.

Copy link
Contributor Author

@mihir6692 mihir6692 Nov 15, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I couldn't see any benefit for Get path.

get = new Get(startKey);
get.setCacheBlocks(false);
get.setFilter(new FirstKeyOnlyFilter());
get.addFamily(column.getName());
} else {
scan = new Scan();
LOG.debug("rawScan {} for {}", rawScanEnabled, region.getTable());
scan.setRaw(rawScanEnabled);
scan.setCaching(1);
scan.setCacheBlocks(false);
scan.setFilter(new FirstKeyOnlyFilter());
scan.addFamily(column.getName());
scan.setMaxResultSize(1L);
scan.setOneRowLimit();
// There are 4 types of region for any table.
// 1. Start and End key are empty. (Table with Single region)
// 2. Start key is empty. (First region of the table)
// 3. End key is empty. (Last region of the table)
// 4. Region with Start & End key. (All the regions between first & last region of the
// table.)
//
// Since Scan only takes Start and/or End Row and doesn't accept the region ID,
// we set the start row when Regions are of type 3 OR 4 as mentioned above.
// For type 1 and 2, We don't need to set this option.
scan.withStartRow(startKey);
}
LOG.debug("rawScan {} for {}", rawScanEnabled, region.getTable());
scan.setRaw(rawScanEnabled);
scan.setCaching(1);
scan.setCacheBlocks(false);
scan.setFilter(new FirstKeyOnlyFilter());
scan.addFamily(column.getName());
scan.setMaxResultSize(1L);
scan.setOneRowLimit();
LOG.debug("Reading from {} {} {} {}", region.getTable(), region.getRegionNameAsString(),
column.getNameAsString(), Bytes.toStringBinary(startKey));
try {
stopWatch.start();
if (startKey.length > 0) {
table.get(get);
} else {
rs = table.getScanner(scan);
rs.next();
}
rs = table.getScanner(scan);
rs.next();
stopWatch.stop();
this.readWriteLatency.add(stopWatch.getTime());
sink.publishReadTiming(serverName, region, column, stopWatch.getTime());
Expand Down